PK/PD Model of Pegylated Thrombopoietin Mimetic Peptide in Healthy Subjects: Comparison of Verification Procedures for Assessing Model Predictability.
Mahesh N. Samtani, Juan Jose Perez-Ruixo, Kathryn Brown, Dirk Cerneus, Christopher Molloy
Johnson & Johnson Pharmaceutical Research & Development, Beerse, Belgium; Raritan, NJ; Spring House, PA; and High Wycombe, UK.
Objective: To model the pharmacokinetics (PK) and pharmacodynamics (PD) of a pegylated thrombopoietin mimetic peptide (PEG-TPOm) in healthy subjects and to compare verification procedures for assessing model predictability.
Methods: 30 healthy subjects were randomized to receive single intravenous doses (0.375, 0.75, 1.5, 2.25 or 3 ug/kg) of PEG-TPOm. Plasma concentrations and platelet counts were analyzed using a target mediated drug disposition model and a precursor pool life span system. To verify the precision, stability, and predictability of the models, the final parameter estimates were subjected to internal model verification. The verification consisted of a non-parametric bootstrap [1], a visual predictive check [2], and posterior predictive assessment [3]. The visual predictive check assesses the influence of variability in parameters while the posterior check adds another level of complexity by incorporating the uncertainty in parameter estimates. Posterior predictive check was performed using parameter uncertainty and variability, which were incorporated using 1000 different values of fixed and random effect parameters that were randomly sampled from the bootstrap replicates. Visual predictive check was also performed using the original parameter estimates. The 5th, 50th and 95th percentiles were calculated from the simulated profiles for posterior and visual predictive checks. The prediction intervals were super-imposed on the raw data to allow assessment of model predictability.
Results: The terminal PK half-life ranged between 18-36 hours and the volume of distribution (5 L) suggested that PEG-TPOm was primarily restricted to the blood compartment. The estimated equilibrium dissociation constant (154 pM) indicated that PEG-TPOm has high affinity for TPO receptors. The megakaryocyte production rate displayed a maximal stimulatory increase of 52% and 22% receptor occupancy was necessary to achieve half maximal stimulatory effect. The increase in platelet counts was observed after a delay of 4 days and maximum platelets counts were achieved on day 9, which is reflective of the megakaryocyte and platelet life spans.
Non-parametric bootstrap analysis was used to validate the model parameter estimates. The original estimates were similar to the median value obtained from the bootstrap analysis and fell within the 90% confidence interval. An overlay of the observed data, the visual predictive check, and the posterior predictive results confirmed that the models are able to capture the majority of PK/PD observations, which fell within the 90% prediction intervals. The results of the visual and posterior predictive checks were in very close agreement with one another. The concordance of the two verification procedures indicated that model uncertainty does not significantly affect model predictability.
Conclusion: A suitable mechanistic model was validated for describing the complex PK/PD of PEG-TPOm in humans, including target-mediated disposition, platelet stimulation and mean life spans of thrombopoietic cell populations. The two model verification procedures involving posterior and visual predictive checks gave similar results indicating that for this PK/PD dataset both methods are suitable for model validation since the predictions are primarily driven by inter-individual variability.
References:
[1] Parke J, Holford NH, Charles BG. A procedure for generating bootstrap samples for the validation of nonlinear mixed-effects population models. Comput Methods Programs Biomed. 1999; 59:19-29.
[2] Holford N. The Visual Predictive Check - Superiority to Standard Diagnostic (Rorschach) Plots. 14th meeting of the Population Approach Group in Europe; 2005 Jun 16-17; Pamplona. Available from URL: http://www.page-meeting.org/default.asp?abstract=738 [Accessed 2007 March 13].
[3] Yano Y, Beal SL, Sheiner LB. Evaluating pharmacokinetic/pharmacodynamic models using the posterior predictive check. J Pharmacokinet Pharmacodyn. 2001; 28:171-92.
